How I Built a Legal Intake and Triage App with Claude Code and Momen Backend

This showcase, built for the Cambridge Hack the Law hackathon, presents Aequitas — a legal-aid intake assistant. Users describe their situation in plain language, attach supporting media (notice photos, screenshots, PDFs, or videos), and the system evaluates urgency, identifies missing information, and suggests next steps or referrals.

The technical architecture pairs a visual Momen backend with a Claude Code frontend deployed to Vercel. Momen BaaS is a visual, Postgres-native backend that AI coding tools and no-code builders plug into: you configure your data model, logic, AI agents, and permissions in the editor, and it exposes a standard GraphQL API your frontend consumes. One backend, many frontends.

Live Demo: hack-the-law-multimodel-intake.vercel.app

Narrative and Evidence In, Structured Triage Out

Legal intake inherently involves multiple input formats. The Aequitas pipeline follows this workflow:

• Free-text narrative with optional contact information

• Multimodal attachments (images, documents, videos)

• Asynchronous AI triage producing structured JSON

• Persistent intake and assessment records

• Referral suggestions filtered from an organization directory

The backend handles "media columns, structured-output agent, async orchestration, reference tables." Claude Code manages the UI layer, reading backend schema through the Momen plugin rather than relying on manual documentation.

What the System Does

App Features

• Public intake form with optional account sign-in

• Free-text narrative field plus optional name and email

• Multimodal attachments: photos/screenshots (IMAGE), documents/PDFs (FILE), short video (VIDEO)

• Asynchronous AI triage returning structured assessment data

• Urgency classification (critical | high | medium | low) with reasoning and deadline

• Missing-facts checklist and recommended next steps

• Referral category assignment for routing

• Confidence scoring on assessments

• No external payment systems or APIs beyond Momen's built-in AI and file storage

Data Model

Five tables comprise the backend architecture:

Assessment columns directly mirror the agent's JSON schema: issue_category, issue_summary, document_type, jurisdiction, parties, key_dates, urgency_level, urgency_reason, deadline, missing_facts, recommended_steps, referral_category, and confidence.

AI Configuration

One multimodal agent functions as a civil legal-aid intake specialist:

Inputs: narrative (TEXT), document_image (IMAGE), document_file (FILE), document_video (VIDEO)

Role: Analyze narrative and supporting documents to triage legal issues, classify them, assess urgency, identify missing information, and suggest referral categories.

Structured Output: issue_category, issue_summary, parties, urgency_level (exactly one of critical | high | medium | low), urgency_reason, deadline, missing_facts, recommended_steps, referral_category, confidence (0.0–1.0)

Optional Output: document_type, jurisdiction, key_dates

All four input types can be passed in a single agent call — no separate OCR or transcription pipeline.

Backend Logic

One async Actionflow named triage_intake orchestrates the process:

1. Receive narrative, attachment IDs, and optional contact fields

2. Insert intake row with pending status

3. Start AI conversation with the triage agent, passing narrative and attached media

4. Insert assessment row, mapping agent JSON fields into typed columns linked to the intake

5. Update intake.status to triaged

6. Return intake_id

The flow runs asynchronously because multimodal model inference exceeds synchronous timeout limits.

Frontend Invocation Pattern:

1. Request presigned URLs for each attachment type → HTTP PUT binary data → collect asset IDs

2. Invoke triage_intake via fz_create_action_flow_task

3. Subscribe via fz_listen_action_flow_result WebSocket until COMPLETED

4. Query intake_by_pk with nested assessment data; filter referral rows by category and jurisdiction

Form Screenshot

Momen BaaS Integration with Claude Code

Momen BaaS exposes your entire visual backend — tables, agents, Actionflows, permissions — as a typed GraphQL API. The Momen plugin gives Claude Code direct access to your backend schema, so it can generate correct frontend code without manual API documentation.

Backend Configuration (Momen Editor):

• Create intake, assessment, and reference tables with relationships

• Seed urgency_level, referral_category, and referral rows

• Configure multimodal triage agent with structured output schema

• Build the triage_intake Actionflow

Frontend Development (Claude Code + Momen BaaS):

Install the Momen plugin:

# Claude Code
claude plugin marketplace add momen-tech-org/momen-nocode-plugin
claude plugin install momen-nocode@momen

Once installed, Claude Code reads your agent inputs, Actionflow names, and output schemas directly. Then:

• Generate intake form UI with multi-type file pickers

• Implement binary upload helpers and async invocation with WebSocket subscription

• Build triage result view and referral list filtered by assessment category

• Deploy to Vercel

Design Approach

• Backend: Headless (no Momen canvas UI)

• Frontend: Three-step hero (Describe → AI triages → Get matched referrals), async processing state, structured triage card, referral suggestions

• Permissions: Open anonymous access for demo; production would use role-based permissions

Technical Highlights

• Small but complete backend with two core business tables and three reference tables

• Multimodal processing in one agent call without separate OCR pipeline

• Structured output mapped 1:1 from agent JSON to Postgres columns

• Async-by-default for AI using task + subscription pattern

• Referral routing via plain table data without custom code

• Visual logic collocated with Postgres in Actionflow, not at a distant Edge layer

Timeline and Cost

Momen Pro is required for multimodal AI agents. Claude Code uses an existing subscription. Vercel free tier hosts the demo. Each triage run consumes AI points for the multimodal model call.

How to Try It

Demo: hack-the-law-multimodel-intake.vercel.app

1. Describe a legal situation in the narrative field

2. Optionally attach a notice photo, document, or short video

3. Submit for triage and wait for the structured assessment

4. Review urgency, missing facts, recommended steps, and referral suggestions

To rebuild a similar backend: create a Momen project, configure the data model and agent visually, install the Momen plugin, and prompt Claude Code to build the frontend against your project schema.

Closing Thoughts

Aequitas demonstrates a repeatable pattern for intake and triage workflows: a visual Momen data model with reference rubrics, one structured-output agent, and one async Actionflow comprise the backend. Claude Code manages the UI. "No custom server routes, no separate object storage, no Edge Function glue." The entire system — from editor configuration to live deployment — highlights how visual backend configuration combined with AI coding tools accelerates development.

Key Links:

• Momen BaaS / Headless Documentation

• Momen Plugin (GitHub)

• File Management Documentation

• AI Agent Documentation

• Database Configuration Guide

• Role-Based Permissions

• Get Started — It's Free